Robust Audio-Visual Speech Recognition under Noisy Audio-Video Conditions

This paper presents the maximum weighted stream posterior (MWSP) model as a robust and efficient stream integration method for audio-visual speech recognition in environments, where the audio or video streams may be subjected to unknown and time-varying corruption. A significant advantage of MWSP is that it does not require any specific measurements of the signal in either stream to calculate appropriate stream weights during recognition, and as such it is modality-independent. This also means that MWSP complements and can be used alongside many of the other approaches that have been proposed in the literature for this problem. For evaluation we used the large XM2VTS database for speaker-independent audio-visual speech recognition. The extensive tests include both clean and corrupted utterances with corruption added in either/both the video and audio streams using a variety of types (e.g., MPEG-4 video compression) and levels of noise. The experiments show that this approach gives excellent performance in comparison to another well-known dynamic stream weighting approach and also compared to any fixed-weighted integration approach in both clean conditions or when noise is added to either stream. Furthermore, our experiments show that the MWSP approach dynamically selects suitable integration weights on a frame-by-frame basis according to the level of noise in the streams and also according to the naturally fluctuating relative reliability of the modalities even in clean conditions. The MWSP approach is shown to maintain robust recognition performance in all tested conditions, while requiring no prior knowledge about the type or level of noise.

Problèmes de perception de la parole dans le bruit chez les enfants présentant un trouble de traitement auditif

Il est bien connu que les enfants qui présentent un trouble de traitement auditif (TTA) ont de la difficulté à percevoir la parole en présence de bruit de fond. Cependant, il n’existe aucun consensus quant à l’origine de ces difficultés d’écoute. Ce programme de recherche est consacré à l’étude des incapacités sous-jacentes aux problèmes de perception de la parole dans le bruit chez les enfants présentant un TTA. Le Test de Phrases dans le Bruit (TPB) a été développé afin d’examiner si les difficultés de perception de la parole dans le bruit d’enfants ayant un TTA relèvent d’incapacités auditives, d’incapacités cognitivo-linguistiques ou des deux à la fois. Il comprend cinq listes de 40 phrases, composées de 20 phrases hautement prévisibles (HP) et de 20 phrases faiblement prévisibles (FP), de même qu’un bruit de verbiage. Le niveau de connaissance du mot clé (mot final) de chaque phrase a été vérifié auprès d’un groupe d’enfants âgés entre 5 et 7 ans. De plus, le degré d’intelligibilité des phrases dans le bruit et le niveau de prévisibilité ont été mesurées auprès d’adultes pour assurer l’équivalence entre les listes. Enfin, le TPB a été testé auprès d’un groupe de 15 adultes et d’un groupe de 69 enfants sans trouble auditif avant de l’administrer à des enfants ayant un TTA. Pour répondre à l’objectif général du programme de recherche, dix enfants présentant un TTA (groupe TTA) et dix enfants jumelés selon le genre et l’âge sans difficulté auditive (groupe témoin) ont été soumis aux listes de phrases du TPB selon différentes conditions sonores. Le groupe TTA a obtenu des performances significativement plus faibles comparativement au groupe témoin à la tâche de reconnaissance du mot final des phrases présentées en même temps qu’un bruit de verbiage compétitif, aux rapports signal-sur-bruit de 0, +3 et +4 dB. La moyenne de la différence des scores obtenue entre les phrases HP et FP à chaque condition expérimentale de bruit était similaire entre les deux groupes. Ces résultats suggèrent que les enfants ayant un TTA ne se distinguent pas des enfants du groupe témoin au plan de la compétence cognitivo-linguistique. L’origine des difficultés d’écoute de la parole dans le bruit dans le cas de TTA serait de nature auditive. Toutefois, les résultats des analyses de groupe diffèrent de ceux des analyses individuelles. Les divers profils de difficultés d’écoute identifiés auprès de cette cohorte appuient l’importance de continuer les investigations afin de mieux comprendre l’origine des problèmes de perception de la parole dans le bruit dans le cas de TTA. En connaissant mieux la nature de ces difficultés, il sera possible d’identifier les stratégies d’intervention de réadaptation spécifiques et efficaces.

L'effort associé à la reconnaissance de la parole chez les adultes et les personnes aînées

L’objectif principal de cette thèse était de quantifier et comparer l’effort requis pour reconnaître la parole dans le bruit chez les jeunes adultes et les personnes aînées ayant une audition normale et une acuité visuelle normale (avec ou sans lentille de correction de la vue). L’effort associé à la perception de la parole est lié aux ressources attentionnelles et cognitives requises pour comprendre la parole. La première étude (Expérience 1) avait pour but d’évaluer l’effort associé à la reconnaissance auditive de la parole (entendre un locuteur), tandis que la deuxième étude (Expérience 2) avait comme but d’évaluer l’effort associé à la reconnaissance auditivo-visuelle de la parole (entendre et voir le visage d’un locuteur). L’effort fut mesuré de deux façons différentes. D’abord par une approche comportementale faisant appel à un paradigme expérimental nommé double tâche. Il s’agissait d’une tâche de reconnaissance de mot jumelée à une tâche de reconnaissance de patrons vibro-tactiles. De plus, l’effort fut quantifié à l’aide d’un questionnaire demandant aux participants de coter l’effort associé aux tâches comportementales. Les deux mesures d’effort furent utilisées dans deux conditions expérimentales différentes : 1) niveau équivalent – c'est-à-dire lorsque le niveau du bruit masquant la parole était le même pour tous les participants et, 2) performance équivalente – c'est-à-dire lorsque le niveau du bruit fut ajusté afin que les performances à la tâche de reconnaissance de mots soient identiques pour les deux groupes de participant. Les niveaux de performance obtenus pour la tâche vibro-tactile ont révélé que les personnes aînées fournissent plus d’effort que les jeunes adultes pour les deux conditions expérimentales, et ce, quelle que soit la modalité perceptuelle dans laquelle les stimuli de la parole sont présentés (c.-à.-d., auditive seulement ou auditivo-visuelle). Globalement, le ‘coût’ associé aux performances de la tâche vibro-tactile était au plus élevé pour les personnes aînées lorsque la parole était présentée en modalité auditivo-visuelle. Alors que les indices visuels peuvent améliorer la reconnaissance auditivo-visuelle de la parole, nos résultats suggèrent qu’ils peuvent aussi créer une charge additionnelle sur les ressources utilisées pour traiter l’information. Cette charge additionnelle a des conséquences néfastes sur les performances aux tâches de reconnaissance de mots et de patrons vibro-tactiles lorsque celles-ci sont effectuées sous des conditions de double tâche. Conformément aux études antérieures, les coefficients de corrélations effectuées à partir des données de l’Expérience 1 et de l’Expérience 2 soutiennent la notion que les mesures comportementales de double tâche et les réponses aux questionnaires évaluent différentes dimensions de l’effort associé à la reconnaissance de la parole. Comme l’effort associé à la perception de la parole repose sur des facteurs auditifs et cognitifs, une troisième étude fut complétée afin d’explorer si la mémoire auditive de travail contribue à expliquer la variance dans les données portant sur l’effort associé à la perception de la parole. De plus, ces analyses ont permis de comparer les patrons de réponses obtenues pour ces deux facteurs après des jeunes adultes et des personnes aînées. Pour les jeunes adultes, les résultats d’une analyse de régression séquentielle ont démontré qu’une mesure de la capacité auditive (taille de l’empan) était reliée à l’effort, tandis qu’une mesure du traitement auditif (rappel alphabétique) était reliée à la précision avec laquelle les mots étaient reconnus lorsqu’ils étaient présentés sous les conditions de double tâche. Cependant, ces mêmes relations n’étaient pas présentes dans les données obtenues pour le groupe de personnes aînées ni dans les données obtenues lorsque les tâches de reconnaissance de la parole étaient effectuées en modalité auditivo-visuelle. D’autres études sont nécessaires pour identifier les facteurs cognitifs qui sous-tendent l’effort associé à la perception de la parole, et ce, particulièrement chez les personnes aînées.

Visual Speech Synthesis by Morphing Visemes

We present MikeTalk, a text-to-audiovisual speech synthesizer which converts input text into an audiovisual speech stream. MikeTalk is built using visemes, which are a small set of images spanning a large range of mouth shapes. The visemes are acquired from a recorded visual corpus of a human subject which is specifically designed to elicit one instantiation of each viseme. Using optical flow methods, correspondence from every viseme to every other viseme is computed automatically. By morphing along this correspondence, a smooth transition between viseme images may be generated. A complete visual utterance is constructed by concatenating viseme transitions. Finally, phoneme and timing information extracted from a text-to-speech synthesizer is exploited to determine which viseme transitions to use, and the rate at which the morphing process should occur. In this manner, we are able to synchronize the visual speech stream with the audio speech stream, and hence give the impression of a photorealistic talking face.

Evaluation of sentence perception of adults with hearing loss when using a video teleconferencing system

This study evaluate the speech perception of hearing-impaired adults (with varying degrees of deafness) when using a video teleconferencing system (an integrated service digital network).

The neural processing of masked speech: Evidence for different mechanisms in the left and right temporal lobes

It has been previously demonstrated that extensive activation in the dorsolateral temporal lobes associated with masking a speech target with a speech masker, consistent with the hypothesis that competition for central auditory processes is an important factor in informational masking. Here, masking from speech and two additional maskers derived from the original speech were investigated. One of these is spectrally rotated speech, which is unintelligible and has a similar (inverted) spectrotemporal profile to speech. The authors also controlled for the possibility of “glimpsing” of the target signal during modulated masking sounds by using speech-modulated noise as a masker in a baseline condition. Functional imaging results reveal that masking speech with speech leads to bilateral superior temporal gyrus (STG) activation relative to a speech-in-noise baseline, while masking speech with spectrally rotated speech leads solely to right STG activation relative to the baseline. This result is discussed in terms of hemispheric asymmetries for speech perception, and interpreted as showing that masking effects can arise through two parallel neural systems, in the left and right temporal lobes. This has implications for the competition for resources caused by speech and rotated speech maskers, and may illuminate some of the mechanisms involved in informational masking.

The neural processing of masked speech: evidence for different mechanisms in the left and right temporal lobes

It has been previously demonstrated that extensive activation in the dorsolateral temporal lobes associated with masking a speech target with a speech masker, consistent with the hypothesis that competition for central auditory processes is an important factor in informational masking. Here, masking from speech and two additional maskers derived from the original speech were investigated. One of these is spectrally rotated speech, which is unintelligible and has a similar (inverted) spectrotemporal profile to speech. The authors also controlled for the possibility of "glimpsing" of the target signal during modulated masking sounds by using speech-modulated noise as a masker in a baseline condition. Functional imaging results reveal that masking speech with speech leads to bilateral superior temporal gyrus (STG) activation relative to a speech-in-noise baseline, while masking speech with spectrally rotated speech leads solely to right STG activation relative to the baseline. This result is discussed in terms of hemispheric asymmetries for speech perception, and interpreted as showing that masking effects can arise through two parallel neural systems, in the left and right temporal lobes. This has implications for the competition for resources caused by speech and rotated speech maskers, and may illuminate some of the mechanisms involved in informational masking.

Temporal-envelope constancy of speech in rooms and the perceptual weighting of frequency bands

Three experiments measured constancy in speech perception, using natural-speech messages or noise-band vocoder versions of them. The eight vocoder-bands had equally log-spaced center-frequencies and the shapes of corresponding “auditory” filters. Consequently, the bands had the temporal envelopes that arise in these auditory filters when the speech is played. The “sir” or “stir” test-words were distinguished by degrees of amplitude modulation, and played in the context; “next you’ll get _ to click on.” Listeners identified test-words appropriately, even in the vocoder conditions where the speech had a “noise-like” quality. Constancy was assessed by comparing the identification of test-words with low or high levels of room reflections across conditions where the context had either a low or a high level of reflections. Constancy was obtained with both the natural and the vocoded speech, indicating that the effect arises through temporal-envelope processing. Two further experiments assessed perceptual weighting of the different bands, both in the test word and in the context. The resulting weighting functions both increase monotonically with frequency, following the spectral characteristics of the test-word’s [s]. It is suggested that these two weighting functions are similar because they both come about through the perceptual grouping of the test-word’s bands.

Investigation of auditory processing disorder and language impairment using the speech-evoked auditory brainstem response

This study investigated whether there are differences in the Speech-Evoked Auditory Brainstem Response among children with Typical Development (TD), (Central) Auditory Processing Disorder (C) APD, and Language Impairment (LI). The speech-evoked Auditory Brainstem Response was tested in 57 children (ages 6-12). The children were placed into three groups: TD (n = 18), (C)APD (n = 18) and LI (n = 21). Speech-evoked ABR were elicited using the five-formant syllable/da/. Three dimensions were defined for analysis, including timing, harmonics, and pitch. A comparative analysis of the responses between the typical development children and children with (C)APD and LI revealed abnormal encoding of the speech acoustic features that are characteristics of speech perception in children with (C)APD and LI, although the two groups differed in their abnormalities. While the children with (C)APD might had a greater difficulty distinguishing stimuli based on timing cues, the children with LI had the additional difficulty of distinguishing speech harmonics, which are important to the identification of speech sounds. These data suggested that an inefficient representation of crucial components of speech sounds may contribute to the difficulties with language processing found in children with LI. Furthermore, these findings may indicate that the neural processes mediated by the auditory brainstem differ among children with auditory processing and speech-language disorders. (C) 2012 Elsevier B.V. All rights reserved.

Visual Speech Segmentation: Using Facial Cues to Locate Word Boundaries in Continuous Speech

Speech is typically a multimodal phenomenon, yet few studies have focused on the exclusive contributions of visual cues to language acquisition. To address this gap, we investigated whether visual prosodic information can facilitate speech segmentation. Previous research has demonstrated that language learners can use lexical stress and pitch cues to segment speech and that learners can extract this information from talking faces. Thus, we created an artificial speech stream that contained minimal segmentation cues and paired it with two synchronous facial displays in which visual prosody was either informative or uninformative for identifying word boundaries. Across three familiarisation conditions (audio stream alone, facial streams alone, and paired audiovisual), learning occurred only when the facial displays were informative to word boundaries, suggesting that facial cues can help learners solve the early challenges of language acquisition.

Statistical Learning, Syllable Processing, and Speech Production in Healthy Hearing and Hearing-Impaired Preschool Children: A Mismatch Negativity Study

OBJECTIVES The objectives of the present study were to investigate temporal/spectral sound-feature processing in preschool children (4 to 7 years old) with peripheral hearing loss compared with age-matched controls. The results verified the presence of statistical learning, which was diminished in children with hearing impairments (HIs), and elucidated possible perceptual mediators of speech production. DESIGN Perception and production of the syllables /ba/, /da/, /ta/, and /na/ were recorded in 13 children with normal hearing and 13 children with HI. Perception was assessed physiologically through event-related potentials (ERPs) recorded by EEG in a multifeature mismatch negativity paradigm and behaviorally through a discrimination task. Temporal and spectral features of the ERPs during speech perception were analyzed, and speech production was quantitatively evaluated using speech motor maximum performance tasks. RESULTS Proximal to stimulus onset, children with HI displayed a difference in map topography, indicating diminished statistical learning. In later ERP components, children with HI exhibited reduced amplitudes in the N2 and early parts of the late disciminative negativity components specifically, which are associated with temporal and spectral control mechanisms. Abnormalities of speech perception were only subtly reflected in speech production, as the lone difference found in speech production studies was a mild delay in regulating speech intensity. CONCLUSIONS In addition to previously reported deficits of sound-feature discriminations, the present study results reflect diminished statistical learning in children with HI, which plays an early and important, but so far neglected, role in phonological processing. Furthermore, the lack of corresponding behavioral abnormalities in speech production implies that impaired perceptual capacities do not necessarily translate into productive deficits.

Children's auditory perception and cognitive processing skills in adverse listening situations

Thesis (Ph.D.)--University of Washington, 2016-06

Visual-auditory integration during speech imitation in autism

Children with autistic spectrum disorder (ASD) may have poor audio-visual integration, possibly reflecting dysfunctional 'mirror neuron' systems which have been hypothesised to be at the core of the condition. In the present study, a computer program, utilizing speech synthesizer software and a 'virtual' head (Baldi), delivered speech stimuli for identification in auditory, visual or bimodal conditions. Children with ASD were poorer than controls at recognizing stimuli in the unimodal conditions, but once performance on this measure was controlled for, no group difference was found in the bimodal condition. A group of participants with ASD were also trained to develop their speech-reading ability. Training improved visual accuracy and this also improved the children's ability to utilize visual information in their processing of speech. Overall results were compared to predictions from mathematical models based on integration and non-integration, and were most consistent with the integration model. We conclude that, whilst they are less accurate in recognizing stimuli in the unimodal condition, children with ASD show normal integration of visual and auditory speech stimuli. Given that training in recognition of visual speech was effective, children with ASD may benefit from multi-modal approaches in imitative therapy and language training. (C) 2004 Elsevier Ltd. All rights reserved.

Measuring the ability of school children with a history of otitis media to understand everyday speech

The present study compared the ability of school-aged children with and without a history of otitis media (OM) to understand everyday speech in noise using the University of Queensland Understanding of Everyday Speech Test (UQUEST). Participants were 484 children (246 boys, 238 girls) attending Grade 3 (272, mean age = 8.25 yr., SD = 0.43) and Grade 4 (212, mean age = 9.28 yr., SD = 0.41) at 19 primary schools in Brisbane metropolitan and Sunshine Coast schools. Children selected for inclusion were native speakers of English with normal hearing on the day of testing and had no reported physical or behavioral impairments. The children were divided into three groups according to the number of episodes of OM since birth. The results showed no significant differences in speech scores across the participant groups. However, a significant difference in mean speech scores was found across the grades and the noise conditions. Although children with a history of OM performed equally well at a group level when compared to the controls, they exhibited a large range of abilities in speech comprehension within the same group.

Cross-language Perception of Word-final Stops by Native Vietnamese Listeners: Preliminary Results on the Role of Specific, Non-native Phonetic Experience

This study examined the discrimination of word-final stop contrasts (/p/-/t/, /p/-/k/, /t/-/k/) in English and Thai by 12 listeners who speak Vietnamese as their first language (L1). Vietnamese shares specific phonetic realization of stops with Thai, i.e., unreleased final stop and differs from English which allows both released and unreleased final stops. These 12 native Vietnamese (NV) listeners’ discrimination accuracy was compared to that of the two listener groups (Australian English (AE), native Thai (NT)) tested in previous studies. The NV group was less accurate than the native group in discriminating both English and Thai stop contrasts. In particular, for the Thai /t/-/k/ contrast, they were significantly less accurate than the AE listeners. The present findings suggest that experience with specific (i.e., unreleased) and native phonetic realization of sounds may be essential in accurate discrimination of final stop contrasts. The effect of L1 dialect on cross-language speech perception is discussed.